STM32WB55

[zyc]

Dear ST Support Team and Community Experts,

I'm developing a BLE application using the STM32WB55 Nucleo board (P-NUCLEO-WB55) and observed unexpectedly low data throughput, requesting guidance on optimization.

1. Issue Description

• Current Rate:
  Modified BLE_p2pServer example to send sensor data via P2PS_Send_notification() in a loop.
  Measured: ~4.5KB/s (7.5ms connection interval).

• Comparative Tests:

  • Enabled L2CAP CoC in Data Throughput example (SW2 press): ~40KB/s (still below expectations).

  • Competing device (nRF52840) achieves >100KB/s under identical conditions (2M PHY + DLE).

• Anomaly:
  No throughput improvement after forcing 2M PHY via aci_hal_set_phy_parameters().

2. Test Environment

• Hardware: STM32WB55CGU6 (Nucleo-68 module).

• Software:

  • Wireless Stack: v1.18.03 (FUS not upgraded).

  • IDE: STM32CubeIDE v1.15.0.

  • Configurations applied per AN5289 Rev 8 (Sections 7.8 & 8.6.9 for DLE/MTU/CoC).

• Connection Parameters:

  gap_role_Init_Parameters.Connection_Interval = 8;

  aci_gatt_exchange_config(connection_handle, 247);

  aci_le_set_data_length(connection_handle, 251, 2120);

3. Attempted Solutions

• Verified successful PHY switching (no HCI errors).

• Tested connection intervals (7.5ms/15ms/30ms) – negligible impact.

• Compared GATT Notify vs. L2CAP CoC: CoC capped at 40KB/s (expected ~150KB/s).

• Upgraded STM32CubeProgrammer to v2.20 (FUS compatibility confirmed).

4. Key Questions

Kindly advise on:

1. Theoretical Limits:
   What is the maximum achievable throughput for STM32WB55 using 2M PHY + DLE + CoC? Are official test reports available?

2. Stack Version Dependency:
   Does Stack v1.18.03 limit throughput? Is upgrading to v1.19+ mandatory (e.g., for EATT optimizations per Release Notes ID 205915)?

3. Configuration Gaps:

   • Is explicit HCI_LE_Set_Phy required? Does RF clock accuracy affect 2M PHY?

   • How to optimize CoC’s Credit-Based Flow Control (e.g., MTU/MPS in aci_l2cap_coc_connect)?

4. vs. nRF52840 Discrepancy:
   Why does WB55 underperform identically configured nRF52840? Is stack overhead higher? Mitigation strategies?

5. Hardware Limitations:
   Could WB55’s 6dBm output power impact 2M PHY stability? Is RF matching circuit tuning needed?

5. Support Requests

• Provide an official throughput-optimized example (full PHY+DLE+CoC implementation).

• Recommend upgrading to latest stack (e.g., v2.x)? Any backward compatibility risks?